Abstract

E-commerce companies such as Amazon, Alibaba and Flipkart process billions oforders every year. However, these orders represent only a small fraction of allplausible orders. Exploring the space of all plausible orders could help usbetter understand the relationships between the various entities in ane-commerce ecosystem, namely the customers and the products they purchase. Inthis paper, we propose a Generative Adversarial Network (GAN) for orders madein e-commerce websites. Once trained, the generator in the GAN could generateany number of plausible orders. Our contributions include: (a) creating a denseand low-dimensional representation of e-commerce orders, (b) train anecommerceGAN (ecGAN) with real orders to show the feasibility of the proposedparadigm, and (c) train an ecommerce-conditional-GAN (ec^2GAN) to generate theplausible orders involving a particular product. We propose several qualitativemethods to evaluate ecGAN and demonstrate its effectiveness. The ec^2GAN isused for various kinds of characterization of possible orders involving aproduct that has just been introduced into the e-commerce system. The proposedapproach ec^2GAN performs significantly better than the baseline in most of thescenarios.

 

Introduction (beta)

None

 

Conclusion (beta)

None